Myeloid cell‐specific mutation of Spi1 selectively reduces M2‐biased macrophage numbers in skeletal muscle, reduces age‐related muscle fibrosis and prevents sarcopenia

Abstract Intramuscular macrophages play key regulatory roles in determining the response of skeletal muscle to injury and disease. Recent investigations showed that the numbers and phenotype of intramuscular macrophages change during aging, suggesting that those changes could influence the aging process. We tested that hypothesis by generating a mouse model that harbors a myeloid cell‐specific mutation of Spi1, which is a transcription factor that is essential for myeloid cell development. The mutation reduced the numbers of macrophages biased to the CD163+/CD206+ M2 phenotype in muscles of aging mice without affecting the numbers of CD68‐expressing macrophages and reduced the expression of transcripts associated with the M2‐biased phenotype. The mutation did not affect the colony‐forming ability or the frequency of specific subpopulations of bone marrow hematopoietic cells and did not affect myeloid/lymphoid cell ratios in peripheral blood leukocyte populations. Cellularity of most myeloid lineage cells was not influenced by the mutation. The Spi1 mutation in bone marrow‐derived macrophages in vitro also did not affect expression of transcripts that indicate the M2‐biased phenotype. Thus, myeloid cell‐targeted mutation of Spi1 influences macrophage phenotype in muscle but did not affect earlier stages of differentiation of cells in the macrophage lineage. The mutation reduced age‐related muscle fibrosis, which is consistent with the reduction of M2‐biased macrophages, and reduced expression of the pro‐fibrotic enzyme arginase. Most importantly, the mutation prevented sarcopenia. Together, our observations indicate that intramuscular, M2‐biased macrophages play significant roles in promoting detrimental, age‐related changes in muscle.


| INTRODUC TI ON
Aging of skeletal muscle causes an inevitable and relentless loss of muscle mass and an increase in muscle fibrosis that reduce function and quality of life. Many of the age-related changes are attributable to senescence of the muscle fibers themselves. For example, old muscle fibers show reductions in protein synthesis and increases in proteolysis that contribute to loss of muscle mass and function during aging (Combaret et al., 2009;Pluskal et al., 1984). In addition, age-related increases in muscle fibrosis are affected by changes in muscle stem cells, called satellite cells, as their myogenic capacity declines and their fibrogenic capacity increases (Brack et al., 2007).
However, these age-related changes in muscle cells are also influenced by aging of non-muscle cells that play key regulatory roles in maintaining normal muscle homeostasis. Because the immune system influences the function of every tissue, aging of the immune system can affect cell functions throughout the body. A fundamental change in the immune system of aging animals occurs at the earliest stages of hematopoietic development.
Differentiating hematopoietic stems cells (HSCs), which give rise to almost all blood cells, experience reduced lymphoid potential and a skewing towards the myeloid lineage (Dorshkind et al., 2020). This age-related, myeloid bias is reflected in the periphery where the numbers of mature myeloid cells increase in the circulation (Della Bella et al., 2007). The elevation in their number coincides with an increase in frailty in the elderly, which may have important influences on muscle health because myeloid cells, especially macrophages, play powerful regulatory roles in muscle growth (Tidball, 2017).
The regulatory influence of macrophages varies according to their activation. Although macrophages exist on a continuous spectrum of functional states (Murray et al., 2014), one end of the spectrum has been designated as the M1 phenotype, which is activated by proinflammatory molecules; at the other end of the spectrum, M2 macrophages are activated by anti-inflammatory cytokines (Mills et al., 2000). Both phenotypes play beneficial roles in muscle repair and regeneration. M1-biased macrophages can increase proliferation of satellite cells, which contributes to muscle regeneration (Bencze et al., 2012). Macrophages biased toward the M2 phenotype are also pro-regenerative through their release of factors that increase muscle growth (Tonkin et al., 2015;Wehling-Henricks et al., 2018) and the production of factors that can increase connective tissue production, which provides a framework for tissue repair (Mills, 2001).
Our previous investigations indicate that some age-related changes of muscle are attributable to changes in intramuscular macrophages. For example, expression of an Nos1 transgene in murine muscle reduced the age-related accumulation of M2-biased profibrotic macrophages and prevented the age-related accumulation of intramuscular collagens (Wang et al., 2015). Similarly, aging human muscle experiences an accumulation in M2-biased macrophages and increased fibrosis (Csapo et al., 2014;Cui et al., 2019), although other investigators report no increase in intramuscular collagen in aging humans (Haus et al., 2007). In addition, transplantation of young bone marrow cells into adult mice reduced sarcopenia and muscle fibrosis in the recipients as they aged (Wang et al., 2019), implicating aging of the immune system with muscle loss and fibrosis.
Those observations suggest a link between M2-biased macrophages and fibrosis in aging muscle. However, the experimental interventions that were used to affect the numbers, phenotype or age of myeloid cells in aging muscle would have also influenced other cell types in that tissue. For example, immune cells that include CD8+ cytotoxic T-cells (Zhang et al., 2014) and regulatory T-cells (Burzyn et al., 2013;Wang et al., 2015) are also present in skeletal muscle where they can influence myogenesis; the functions of those cells could also be affected in mice receiving heterochronic bone marrow transplantation (BMT).
In the present investigation, we generated a mouse line in which the Spi1 gene is mutated in myeloid lineage cells to determine the regulatory role of Spi1 in myeloid cells that are present in aging muscle. Spi1 encodes the transcription factor PU.1, which is essential in determining the differentiation fate of hematopoietic cells (DeKoter et al., 2007). High levels of expression of Spi1 are required for differentiation of common myeloid progenitors (CMPs) into mature monocytes/macrophages (Lichanska et al., 1999). Although germ line deletion of Spi1 results in almost complete hematopoietic failure and death of mice in utero or within days after birth (McKercher et al., 1996;Scott et al., 1994), we reasoned that macrophagedeficient mice could be generated by crossing Spi1 floxed mice with LysM Cre mice, in which Cre recombinase is driven by the promoter of the lysozyme2 gene (lyz2) that is expressed exclusively in myeloid cells. A previous study showed that crossing LysM Cre mice with mice with loxP-flanked target genes results in high deletion efficiency of target genes in mature macrophages (Clausen et al., 1999).
Our findings reveal unexpected effects of lyz2-driven deletion of Spi1. We found that the mutation did not reduce the numbers of differentiated, CD68+ macrophages in aging muscle; instead the mutation reduced the numbers of intramuscular macrophages that were activated to the CD163+/CD206+, M2-biased phenotype. This outcome provided us with a tool for assessing the role of M2-biased macrophages in muscle aging and allowed us to validate the important role of that macrophage phenotype in sarcopenia and fibrosis of aging muscle.
2 | RE SULTS 2.1 | Spi1 mutation driven by the lyz2 promoter reduces the number of M2-biased macrophages in muscle LysM Cre /Sfpi1 Lox mice (referred to as Spi1-mutants hereafter) were analyzed up to 22 months of age, which revealed that they exhibited normal survival and no obvious morphological or behavioral differences compared to their LysM wildtype /Sfpi1 Lox littermates (referred to as floxed-controls hereafter). We assayed for Spi1-expressing cells in quadriceps muscles of 22-months-old mice by immunohistochemistry and found that Spi1-mutant mice had more than a 90% reduction in PU.1+ cells compared to age-matched floxed-control mice ( Figure 1a,b).
Because intramuscular macrophages consist of heterogeneous subpopulations (Tidball, 2017), we examined the effect of Spi1 mutation on subpopulations of intramuscular macrophages. Surprisingly, we found that the number of CD68+ macrophages did not differ in quadriceps of Spi1-mutant mice compared to floxedcontrol mice at 12-or 22-months of age (Figure 1c,d) and Cd68 mRNA expression did not differ between floxed-control or Spi1mutant mice at 12 and 22 months of age ( Figure 1e). Those data show that myeloid-specific mutation of Spi1 did not prevent the differentiation of monocytes/macrophages in muscle. However, the numbers of CD163+ M2-biased macrophages were significantly lower in 12-and 22-month-old Spi-mutant mice compared to agematched floxed-control mice (Figure 1f,g). Our QPCR results also showed a trend for reduced expression of Cd163 in quadriceps of Spi1-mutant mice compared to floxed-control mice at both 12 and 22 months (Figure 1h). Because we found that ~73% of CD68+ macrophages in muscles of floxed-control mice expressed CD163 but only ~27% of CD68+ macrophages in Spi1-mutant mice expressed CD163 ( Figure S1a,b) while Spi1 mutation did not affect total number of CD68+ macrophages (Figure 1d), the findings indicate that the mutation reduced the activation of macrophages to a CD163+, M2-biased phenotype without affecting total macrophage numbers in aging muscle.
We also assayed the expression of CD206, another marker of M2-biased macrophages in muscle (Vidal et al., 2008;Villalta et al., 2011;Wang et al., 2014) and found that CD206+ macrophage numbers were reduced by Spi1 mutation in both 12-and 22-month-old muscles (Figure 1i,j). Mrc1 (which encodes CD206), also showed significantly lower expression in Spi1-mutant mice compared to floxed-controls at 12-months of age and a trend for lower expression in mutant mice at 22-months of age ( Figure 1k). Notably, some CD163+ intramuscular macrophages did not express detectible levels of CD206 ( Figure S1c), showing that expression of CD163 and CD206 do not indicate an identical population of M2-biased macrophages in muscle.
Although the decreases in CD163+ and CD206+ cell numbers could reflect impaired macrophage differentiation into the M2 phenotype, they may also reflect the downregulation of CD163 and CD206 expression in M2-biased macrophages because the genes encoding CD163 and CD206 have PU.1 binding sites in their promoter regions and their promoter activity is directly regulated by PU.1 (Eichbaum et al., 1997;Ritter et al., 1999). We tested whether other transcripts that reflect macrophage activation to an M2-biased phenotype that were not direct targets of PU.1 were also affected by the mutation. We first confirmed that CD163+, intramuscular macrophages expressed arginase-1 ( Figure 1l) and then found through QPCR analysis that expression of Arg1 (which encodes arginase-1) was significantly reduced by myeloid-specific mutation of Spi1 in 22-month-old muscles ( Figure 1m). We also observed that expression of Retnla (which encodes Fizz-1, another M2 phenotypic marker) was decreased in 12-and 22-month-old Spi1-mutant mice compared to age-matched floxed-control ( Figure 1n). These data indicate that myeloid cell-specific mutation of Spi1 selectively reduced M2-biased macrophage numbers in muscles at both 12 and 22 months of age.

| Myelopoiesis is intact in Spi1-mutant mice
We assayed whether the Spi1-mutation would disrupt myelopoiesis by testing for effects of the mutation on the proportion of peripheral blood leukocytes that exhibited myeloid cell morphology but found no differences in the proportion of circulating leukocytes that was comprised of myeloid cells in Spi1-mutants and floxed-controls at either 12-or 22-months of age (Figure 2a,b). However, we did observe that aging similarly increased the proportion of peripheral blood leukocytes that were comprised of myeloid cells in both Spi1-mutant and floxed-control mice (Figure 2b).
Because lyz2 is expressed in relatively mature myeloid cells and not progenitors, we expected that differences in primary myelopoiesis would not explain the reduced number of M2-biased macrophages in Spi1 mutants compared to floxed-control mice. We first quantified the number of myeloid colonies generated in semisolid medium by culturing BMCs with a cocktail of myelopoietic cytokines and found no differences in the colony-forming ability of BMCs iso-

| Myeloid cell-specific mutation of Spi1 reduced connective tissue accumulation in old muscle
We showed in a previous investigation that the age-related shift toward greater numbers of CD163+ M2-biased macrophages in muscle is associated with increased muscle fibrosis (Wang et al., 2015). Because Spi1mutant mice showed reduced numbers of M2-biased macrophages in both adult and old muscles compared to age-matched floxed-control mice, we tested whether this reduction in CD163+ and CD206+ macrophages affected muscle fibrosis during aging. Expression of Col1a1 . Furthermore, we observed that myeloid-specific mutation of Spi1 prevented the age-related accumulation of collagen type I and significantly reduced the accumulation of collagen type III during aging (Figure 4g,h). These findings indicate that the increase of M2-biased macrophages during aging contributes to age-related muscle fibrosis and that the myeloid cell-specific Spi1 mutation can reduce the accumulation of connective tissue in aging muscle by decreasing macrophages that are biased toward the M2 phenotype.

| Myeloid cell-specific mutation of Spi1 prevented sarcopenia
We then tested whether the myeloid cell-specific mutation affected sarcopenia by quantifying the cross-sectional area (CSA) of muscle fib-

| Myeloid cell-specific mutation of Spi1 reduced satellite cell activation in adult and aging muscles
We tested whether reducing M2-biased macrophage numbers or activation by myeloid-cell-specific mutation of Spi1 affected satellite cell numbers in aging muscle by assaying for the numbers of muscle cells that expressed the myogenic transcription factors Pax7 and MyoD. Pax7 is expressed by quiescent satellite cells or by recently activated satellite cells that have the potential to return to the reserve population of quiescent satellite cells (Zammit et al., 2006). MyoD is expressed by recently activated satellite cells that have the potential to withdraw from the cell cycle and proceed through terminal differentiation (Smith et al., 1994).
Although the Spi1 mutation did not affect the number of Pax7+

| DISCUSS ION
In the present study, we demonstrate that the number of macrophages in aging muscle that are biased to the CD163+/CD206+ M2 phenotype is selectively and significantly reduced by a myeloid cellspecific deletion of Spi1 without affecting total numbers of CD68+ Importantly, this specific perturbation of M2-biased macrophage numbers was sufficient to reduce sarcopenia and fibrosis of aging muscle.

F I G U R E 4 Myeloid cell specific mutation of
Although previous studies showed that interventions that reduced the numbers of macrophages in aging muscle also produced reductions in connective tissue accumulation (Wang et al., 2015), the experimental approaches used in those studies could have affected non-myeloid cell populations with unknown roles in muscle aging. Our previous findings also showed that the severity of sarcopenia is affected by aging of the immune system (Wang et al., 2019). However, that investigation did not identify which specific, aging immune cells were responsible for increasing sarcopenia. We now show that reducing numbers of intramuscular, M2-biased macrophages significantly reduces degenerative changes that occur in aging muscle. For panels c-f, * indicates significant difference in age-matched groups between genotypes. # indicates significant differences between ages of same genotype. The results of our investigation include several unexpected outcomes. First, the data demonstrate that M2-biased macrophages can exert detrimental effects on homeostasis of non-injured muscle.
This contrasts with the general view that M2-biased macrophages serve beneficial roles in muscle health, based on their demonstrated functions in increasing muscle repair and growth following increased muscle use or acute injury (Tidball & Wehling-Henricks, 2007;Wang et al., 2014); those observations provided the basis for classifying M2-biased macrophages as "reparative" or "pro-regenerative" macrophages. At least in part, the beneficial effects of M2-biased macrophages on muscle repair and regeneration are mediated by deactivating proinflammatory M1-biased macrophages and releasing factors such as IGF1 and Klotho that directly promote myogenesis (Dumont & Frenette, 2010;Tonkin et al., 2015;Wehling-Henricks et al., 2018). However, our current findings show that M2-biased macrophage function in aging muscle is more similar to their role in chronic disease, at least with regard to muscle fibrosis. For example, the chronic muscle wasting disease that occurs in the mdx mouse model of Duchenne muscular dystrophy involves progressive muscle fibrosis that is accompanied by an accumulation of intramuscular M2-biased macrophages that express TGFβ and arginase (Vidal et al., 2008;Villalta et al., 2009)  We were also surprised that myeloid-specific mutation of Spi1 selectively reduced numbers of CD163+ and CD206+ macrophages   (Menezes et al., 2016). Those findings also suggest a link between levels of PU.1 expression and myeloid cell phenotype specification. Furthermore, Spi1+/− mice had reduced numbers of blood CD115 + Ly6C lo cells when presented with inflammatory stimulation and Spi1+/− mice generated iNOS+ macrophages in the spleen more efficiently than wild-type mice (Menezes et al., 2016). These observations are generally consistent with our observations in our Spi1-mutant mice, which indicate that decreasing, but not eliminating, the expression of Spi1 in myeloid cells reduces M2 polarization and promotes M1 polarization of macrophages, without preventing the differentiation of hematopoietic stem cells into the myeloid lineage.
Another particularly intriguing observation in our present study is that the phenotypes of BMDMs in vitro and macrophages in muscles were affected differently by myeloid cell-specific mutation of Spi1. Although intramuscular macrophages showed a specific reduction in M2-biased macrophages, BMDMs from Spi1-mutant mice showed no differences in M2-macrophage-related genes compared to floxed-control BMDMs. These differences in effects of the mutation in muscle and in vitro may be attributable to the different environments in which these macrophages developed and were activated. BMDMs were cultured in vitro with recombinant M-CSF to drive them towards macrophage differentiation, while intramuscular macrophages in aging muscle were activated in a more complex inflammatory environment. An important next step in future investigations will be to determine whether the differences between mechanisms that regulate macrophage phenotype in vitro and in aging muscle in vivo, as we report here, are specific to macrophage activation in the aging muscle microenvironment or regulate macrophage function in multiple, aging tissues throughout the body.
We also observed that the response of Spi1-mutant BMDMs to forced M2 polarization with IL-4 and IL-10 treatment did not differ from the response of floxed-control BMDMs. However, a previous investigation showed that Spi1 +/− BMDMs displayed a blunted response to IL-4 treatment in M2-related gene expression (Qian et al., 2015). A major difference between these two models is that the Spi1 mutation in our model is driven by LysM Cre , whereas Spi1 +/− As with other investigations using mouse models to study complex in vivo mechanisms, the extent to which the findings presented here pertain to the regulatory roles on macrophages in age-related changes in human muscles will need to be tested experimentally.

BMDMs have reduced
The genomic response of human tissues to trauma or disease that involves an inflammatory response shows poor correspondence to the genomic response in mouse models of trauma in which inflammation occurs. For example, changes in gene expression in human tissues experiencing acute trauma, burn, sepsis or infection showed little correlation to changes in gene expression in mouse tissue subjected to the similar, acute inflammation (Seok et al., 2013). The extent to which the murine immune response to muscle aging resembles the human response has not been examined in detail.
Although further investigations are needed to uncover the pro-
Following euthanasia by inhalation of isoflurane, muscles were collected and frozen for sectioning and histological evaluation or frozen in liquid nitrogen until used for RNA isolation. Experimental groups included from 5-8 male mice per group.

| RNA isolation and quantitative PCR
Muscles were homogenized in Trizol (Invitrogen) and RNA extracted, isolated and DNase-treated using RNeasy spin columns according to the manufacturer's protocol (Qiagen). RNA was electrophoresed on 1.2% agarose gels and RNA quality assessed by determining 28S and 18S ribosomal RNA integrity. Total RNA was reverse transcribed with Super Script Reverse Transcriptase II using oligo dTs to prime extension (Invitrogen) to produce cDNA. The cDNA was used to measure the expression of selected transcripts using SYBR green qPCR master mix according to the manufacturer's protocol (Bio-Rad). Real-time PCR was performed on an iCycler thermocycler system equipped with iQ5 optical system software (Bio-Rad). Reference genes were chosen following previously described methods (Wang et al., 2015). Based on that analysis, Rps4x and Srp14 were used as reference genes for QPCR experiments. The normalization factor for each sample was calculated by geometric averaging of the Ct values of both reference genes using the geNorm software. Primers used for QPCR are listed in Table S1.

| Immunohistochemistry and quantification of positive cells
Frozen cross-sections were cut from the mid-belly of quadriceps or TAs at a thickness of 10 μm. The sections were air-dried for 30 min and fixed in ice-cold acetone for 10 min, and endogenous peroxidase ac- . Antibodies to Pax7 were isolated from the conditioned medium as described previously (Wang et al., 2015).

| Peripheral blood leukocytes counting
Whole blood was collected from a femoral bleed. Red blood cells were lysed with ACK lysis buffer (Biowhitaker) that had been precooled on ice for 5 min. Blood samples were washed with Dulbecco's phosphate buffered saline (DPBS; Sigma) and then centrifuged for 5 min in clinical centrifuge at 1000 rpm. Pelleted cells were resuspended in 1 ml of DPBS and 200 μl aliquots of resuspended cells were centrifuged onto microscope slides using a Shandon 3 cytofuge for 5 min at 380 rpm.
Samples were then rinsed briefly in DPBS, fixed with 2% formaldehyde solution for 5 min and stained with hematoxylin for 10 min.
Slides were rinsed with distilled water and cell counts were performed using standard morphological criteria (O'Connell et al., 2015).

| Flow cytometry
BMCs were isolated as described above. Cells were then incubated in anti-CD16/CD32 for 10 min to block Fc receptor binding. HSCs were then identified using combinations of antibodies conjugated to FITC, PE, PerCP/cy5.5 or Pacific Blue, as described previously (Table S2;
BMDMs were then stimulated for 6-h with activation media consisting of DMEM with 0.25% heat-inactivated FBS, penicillin, streptomycin and 10 ng/ml M-CSF with or without IL-4 (25 ng/ml; Sigma) and IL-10 (10 ng/ml; Sigma). RNA was collected in Trizol for QPCR analysis as described above. PPIA and TPT1 were used as reference genes for QPCR experiments using BMDMs.
BMDMs were then stimulated for 24 h with activation media consisting of DMEM with 0.25% heat-inactivated FBS, penicillin, streptomycin, and 10 ng/ml M-CSF. Coverslips were then fixed with 4% paraformaldehyde for 10 min, washed with PBS, and blocked in 3% BSA and 2% gelatin in 50 mM Tris buffer (pH 7.2) for 1 h. Coverslips were then immunolabeled with anti-PU.1 for 3 h. Coverslips labelled with anti-PU.1 were washed with PBS and probed with biotin-conjugated secondary antibodies (Vector) for 30 min. Coverslips were subsequently washed with PBS and incubated with avidin D-conjugated horseradish peroxidase for 30 min. Staining was visualized with the peroxidase substrate 3-amino-9-ethylcarbazole. The number of immunolabeled cells was counted using a bright-field microscope and expressed as the percentage of positive cells per total number of cells.

| Cross-sectional area measurement
Frozen cross-sections were cut from the mid-belly of quadriceps femoris and tibialis anterior (TA) muscles at a thickness of 10 μm.
Sections were then stained with hematoxylin for 10 min. The muscle fiber cross-sectional area was measured for 500 fibers randomly sampled from complete cross-sections using a digital imaging system (Bioquant).

| Statistics
Data are presented as mean ± SEM. One-way analysis of variance was used to test whether differences between 3 or more groups were significant at p < 0.05. Significant differences between groups were identified using Tukey's post hoc test. Comparisons of two groups of values were analyzed using the unpaired, two-tailed t-test.

ACK N OWLED G M ENTS
Research reported in this publication was supported by the U.S.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.